Models, Tools and Behaviours for an Ambient Intelligence Approach to Self-Adaptive Crisis Management - AmICriM

The AmICriM project aims at providing the models, tools and behaviours needed to design distributed embedded applications on electronic devices which we will call Ambient Intelligence Devices (AmID : devices ready to integrate en Ambient Intelligence environment). The general idea of Ambient Intelligence (AmI) is to have multiple heterogeneous, possibly mobile or embedded, electronic devices throughout the environment interacting dynamically so as to satisfy the needs of the users evolving in it. Given the generic nature and complexity of this global goal, we are convinced that we need to provide these AmID with autonomous behaviours enabling them to self-organise to adapt to the environment. In particular, strong dynamics need to be taken into account: devices appearing and disappearing, users changing their needs, ...
The AmICriM project will provide all the software engineering tools (models, guides, behavioural components, middle-ware...) needed to design the software (communication, interaction behaviour, user profiles, data managing...) of AmID and thus the adaptive systems they will constitute. The applicative objective is to obtain a self-adaptive crisis management system taking full advantage of the new pervasive and trusted network and services infrastructures. The main application scenario is the autonomous distributed operational mapping for crisis situations: the possibility to "map" a damaged/critical/unknown zone by using the interaction of numerous ambient devices dispersed in the zone which could provide missing information like localisation of people in danger (and assessment of the danger), specific persons (doctors for example), presence of fire, toxic waste, flooding, electricity or not...

The AmICriM project concerns the rising challenges we find in ambient intelligence, autonomic, pervasive and ubiquitous computing, or domotics. Its ambition is to answer these challenges, among others, in a new way by progressively abandoning finalism when designing artificial systems and strongly relying on cooperative self-organisation mechanisms and emergence. This approach has been already tested on several applications (flood forecast, scheduling, biological modelling...) leading to the AMAS (Adative Multi-Agent Systems) theory and the associated ADELFE (www.irit.fr/ADELFE) toolkit.
This project will have to tackle the following scientific challenges:

• Interoperability between autonomous and heterogeneous (nature, goals and competences) agents evolving in an AmI system: all the possible spectre of electronic devices as well as the humans inside the system, also offering a huge diversity (objectives, limitations).
• The capacity to take into account the limited knowledge during design of the global goal of the system. This limitation is due to the pluri-objective, fuzzy and thus only roughly and incompletely specifiable nature of this goal.
• The fact that the system is situated in a real world and mixed (humans and machines) environment, that it is physically distributed, possessing specific limitations (response time, connectivity) and that it has to endure failures and malfunctioning (graceful degradation).
• The openness of the system with the appearance and disappearance of potentially unknown agents in real time, as well as the adaptation of the agents to their environment.
• The ability to manage the dynamics of the environment and the necessity to react to unexpected situations, and thus to realise truly and continuously adaptive artificial systems.
• The automatic deployment of such systems in real world environments and the scaling up to large number of devices.

Ambient intelligence, self-organisation, adaptive systems, software design

From network and communication to software engineering, methods and models. Also some knowledge on profiling, services management, QoS...
Expertise in Ambient Intelligence, user needs, specific applications, test-beds.

The partners will have to contribute to developing the software engineering tools which will constitute the design plat-form.
Industrials would be expected ton conduct large scale real world experiments. Also provide information on user needs, existing technologies and services.

Both industrials and academics will be needed to have a relevant consortium to tackle the broad spectre of challenges of this project.

Yes